Berlin Chen

- Education

Ph.D. Computer Science and Information Engineering
National Taiwan University           
Dissertation Advisor: Prof. Lin-shan Lee
 
B.A. & M.S. Computer Science and Information Engineering
National Chiao Tung University 
Thesis Advisor: Prof. Chi-Min Liu

- Awards & Honors

  Best Paper Award: Conference on Computational Linguistics and Speech Processing (ROCLING 2016), Taiwan: Paper entitled "Leveraging multi-task learning with neural network based acoustic modeling for improved meeting speech recognition." (in Chinese)
  Best Paper Award: Conference on Computational Linguistics and Speech Processing (ROCLING 2013), Taiwan: Paper entitled "Improved sentence modeling techniques for extractive speech summarization." (in Chinese)
  Excellent Paper Award; The 2012 Conference on Technologies and Applications of Artificial Intelligence (TAAI 2012), Taiwan: Paper entitled "Empirical comparisons of various pseudo-relevant document selection methods for improved spoken document retrieval" (in Chinese)
  Best Paper Award: Conference on Computational Linguistics and Speech Processing (ROCLING 2012), Taiwan: Paper entitled "Recurrent neural network-based language modeling with relevance information" (in Chinese)
  Best Paper Award: Conference on Computational Linguistics and Speech Processing (ROCLING 2011), Taiwan: Paper entitled "Empirical comparisons of various discriminative language models for speech recognition" (in Chinese)
  Advisor, The 11th ACLCLP (中華民國計算語言學學會) Ph.D. Dissertation Award (佳作), Taiwan, 20011 (Student: Mr. Shih-Hsiang Lin): Research entitled "Speech Summarization - Features, Models and Applications"
  Adviser, Best Student Paper Award (Student: Mr. Hung-Shin Lee), 2008 International Symposium on Chinese Spoken Language Processing (ISCSLP2008), Kunming, China: Paper entitled "Improved Linear Discriminant Analysis Considering Empirical Pairwise Classification Error Rates"
Advisor, The 7th ACLCLP (中華民國計算語言學學會) Master Thesis Award (佳作), Taiwan, 2007 (Student: Mr. Shih-Hsiang Lin): Research entitled "Exploring the Use of Data Fitting and Clustering Techniques for Robust Speech Recognition"
Advisor, The 5th ACLCLP (中華民國計算語言學學會) Master Thesis Award (佳作), Taiwan, 2005 (Student: Mr. Jen-Wei Kuo): Research entitled "An Initial Study on Minimum Phone Error Discriminative Learning of Acoustic Models for Mandarin Large Vocabulary Continuous Speech Recognition"
Co-advisor of the Gold-Medal Winner Team at the Fifth TiC100 Contest, Taiwan, 2003
The 10th Acer Dragon (Long-Term) Master Thesis Award (1996): Research entitled "Speaker-Independent Mandarin Polysyllabic Word Recognition", Taiwan, 1996 (Advisor: Prof. Chi-Min Liu)

- Job Experiences

Aug. 2016 ~ Chairman, Department of Computer Science and Information Engineering,
National Taiwan Normal University, Taipei, Taiwan, ROC
 
Feb. 2010 ~ Professor, Department of Computer Science and Information Engineering,
National Taiwan Normal University, Taipei, Taiwan, ROC
 
Feb. 2010 - July 2011 Chief Director, Information Technology Center,
National Taiwan Normal University, Taipei, Taiwan, ROC
 
Aug. 2006 - Jan. 2010 Associate Professor, Department of Computer Science and Information Engineering,
National Taiwan Normal University, Taipei, Taiwan, ROC
 
Aug. 2002 - July 2006 Assistant Professor, Graduate Institute of Computer Science and Information Engineering,
 National Taiwan Normal University, Taipei, Taiwan, ROC
 
Feb. 2002 - July 2002 Adjunct Assistant Professor, Graduate Institute of Electrical Engineering,
National Taipei University of Technology, Taipei, Taiwan, ROC
 
Feb. 2002 - July 2002 Adjunct Assistant Professor, Department of Information Communications,
Shih Hsin University, Taipei, Taiwan, ROC
 
Dec. 2001 - July 2002 Postdoctoral Researcher, Graduate Institute of Communication Engineering,
National Taiwan University, Taipei, Taiwan, ROC
 
Oct 1996 - Nov.  2001 Research Assistant,
Institute of Information Science, Academia Sinica, Taipei, Taiwan, ROC
 
July 1995 - Sept. 1995 Research Assistant (Summer Job), Computer & Communication Research Lab, Industrial Technology Research Institute, Hsinchu, Taiwan, ROC  
July 1992 - Sept. 1992  Technician (Summer Job), Microelectronic Technology Inc, Hsinchu, Taiwan, ROC  
July 1991 - Sept. 1991 Technician (Summer Job), Microelectronic Technology Inc, Hsinchu, Taiwan, ROC  

- Professional Experiences

February 2010 ~   I am a professor of the Computer Science and Information Engineering Department at National Taiwan Normal University. My research interests generally lie in the areas of speech and natural language processing, information retrieval, and artificial intelligence:

1) I and my Ph.D. student proposed a novel summarization framework built on top of notion of risk minimization. This framework can provide a principled way to render the redundancy and coherence relationships among sentences and between sentences and the whole document, as well as leverage the merits of various supervised and unsupervised summarization models.
2
) I and my graduate students proposed a relevance language modeling framework for speech recognition, as well as spoken document retrieval and summarization, which suggests a promising avenue for the integration of relevance, topic and proximity information. The notion of this modeling framework also has been adapted to discriminative language model training.
3) I and Prof. Jeih-weih Hung, as well as my
graduate student, presented a novel use of nonnegative matrix factorization (NMF) and probabilistic latent semantic analysis (PLSA) for deriving noise-robust speech features. The empirical observations showed that the basis spectra constructed on top of NMF and PLSA correspond well with the intuitive notion of the important components (or parts) of modulation frequency.

 

 

Aug. 2006 - Jan. 2010 


 

I was an associate professor with the Department of Computer Science and Information Engineering, National Taiwan Normal University. My research interests included robust feature extraction, acoustic and language modeling, search algorithms for large-vocabulary continuous speech recognition, and speech IR, summarization and mining:

1) I proposed word topic models (WTM), and then worked together with my students to conduct a series of experiments of WTM for language model adaptation in LVCSR.  WTM also has been applied to the speech retrieval and summarization tasks with good success. We also proposed to exploring position information for language modeling in speech recognition. I'm currently working on discriminative training of WTM for IR, initially demonstrated with promising results. This will make WTM more appealing under the existing document ranking criteria.

2) I proposed a novel probabilistic generative framework for extractive spoken document summarization, which can incorporate various sentence generative models and sentence prior models for sentence ranking. I then worked with my students to conduct a series of comprehensive experiments to validate the effectiveness of such a framework. We also continued to conduct a series of experiments on comparing among a number of summarizers and heterogeneous features for spoken document summarization. Several interesting and insightful observations were obtained.

3) I proposed using temporal-spatial feature distribution characteristics for robust speech recognition. I then cooperated with my students to conduct a series of experiments on using such characteristics. We also presented a new histogram equalization approach (called CPHEQ), which demonstrated superiority over a few existing feature normalization approaches for robust speech recognition.

4) I proposed using various levels of training data selection for for discriminative training of acoustic models. I then cooperated with my students to conduct a series of experiments on using such data selection. I also designed a novel frame-level accuracy function for MPE-based discriminative acoustic model training.

5) I and my students developed various improved approaches for applying linear discriminant analysis (LDA) to speech recognition (especially, LVCSR). We extensively explored the relationship between the empirical classification error rates and the Mahalanobis distances of the respective speech phoneme class pairs in the LDA framework.

 

 

Aug. 2002 - July 2006




 

I was an assistant professor of the Graduate Institute of Computer Science and Information Engineering, National Taiwan Normal University. My research interests focused on spoken document retrieval (SDR) and Mandarin large vocabulary speech recognition (LVCSR):

1) I presented a discriminative language modeling (LM) approach for SDR, which might be regarded as the earliest LM-based IR with the learning-to-rank concept, and had demonstrated with very good results and attracted several following efforts. 

2) I designed and implemented a Mandarin LVCSR system with look-ahead features by myself, and had proposed and implemented various document topic models (DTM) for language modeling, and spoken document retrieval, summarization and organization.

3) I also cooperated with my students to implement minimum phone error (MPE) training for Mandarin LVCSR and extended MPE for language model training.

 

 


Dec. 2001 - July 2002 



 
 
 

I was a postdoctoral researcher with the Speech Processing Laboratory at the Department of Electrical Engineering, National Taiwan University.  I instructed both the CS and EE graduate students. Meanwhile, I also served as an adjunct assistant professor at the Graduate Institute of Electrical Engineering, National Taipei University of Technology and the Department of Information Communications, Shih Hsin University, both located in Taipei. I taught Speech Signal Processing and Internet Resource Applications, respectively, in these two Universities.

1)
I proposed a data-driven subword-level indexing approach to Mandarin spoken document retrieval. I then cooperated with Mr. Chun-Jen Wang and Prof. Lin-shan Lee to conduct a series of experiments on using such a indexing approach.

 


Oct. 1996 - Nov. 2001 
 
 
 
 
 
 
 

 

I was a research assistant with the Institute of Information Science, Academia Sinica, Taipei, Taiwan, ROC. My research interests include both the areas of speech recognition, speech-based information retrieval, and Chinese language processing. I had developed several research prototype systems such as the word recognizer, keyword spotter, large vocabulary continuous speech recognizer, and speech-based audio/video information retrieval system, etc. Also, I had joined several research projects sponsored by National Science Council of Republic of China, including “Research on Mandarin Speech Database Retrieval” (Aug 1997 – July 1998), “Research on Mandarin Speech Database Retrieval Using Quasi-Natural-Language Queries” and “Research on Automatic Recognition of Mandarin Broadcast News Speech” (Aug 1999 – July 2000).

 

July 2000 - Aug. 2000
   
 
 

I was a team member of Project MEI (Mandarin-English Information -- Investigating Translingual Speech Retrieval) at the Johns Hopkins University Summer Workshop 2000, sponsored by the National Science Foundation in the US. The team developed one of the first English-Chinese cross-media and cross-language spoken document retrieval systems. English text queries are used to retrieve Mandarin audio documents. MEI is a collaborative project across eight institutions.

 

 

Sept. 1998 - June 2001
    
 
 
 
I was a Ph.D. student with the Digital Speech Processing Laboratory (under the direction of Prof. Lin-shan Lee) at the Department of Computer Science and Information Engineering, National Taiwan University. In my Ph.D. dissertation, I extensively studied the problem of voice retrieval of Mandarin speech information and developed several techniques for content-based audio indexing, term weighting and probabilistic modeling of information retrieval. I also implemented one of the first prototype systems for voice retrieval of Mandarin broadcast news speech collected in Taiwan.

 

Sept. 1994 - June 1996
 
   

 

 I was a master student with the Digital Signal Processing Laboratory (under the direction of Prof. Chi-Min Liu) at the Department of Computer Science and Information Engineering, National Chiao Tung University. In my master thesis, I had successfully implemented a search framework to obtain the heuristic functions for rapid A* search algorithm (with tree-trellis structures) for large vocabulary Mandarin polysyllabic word recognition.

 

 

-  Invited Talks

2016/06/29 Development Center for Compilation and Translation, National Academy for Educational Research (NAER)
    Recent Developments in Machine Learning-based Text Readability Assessment
2016/05/12 Department of Information Management, Yuan Ze University (YZU) 
    Several New Representation Learning Approaches to Automatic Speech Recognition and its Applications
2016/04/20 National Taiwan Normal University (NTNU)
    Mandarin Mispronunciation Detection
2015/10/14 One Day Workshop on Optimization, Department of Mathematics, National Taiwan Normal University (NTNU)
    Recent Developments in Statistical Modeling Techniques for Speech and Natural Language Processing
2015/06/26 Research Center for Psychological and Educational Testing (RCPET), National Taiwan Normal University (NTNU)
    Recent Developments in Automatic Summarization
2014/12/29 CSIE Department, National Taiwan University of Science and Technology (NTUST)
  Title:  Recent Progress in Automatic Speech Recognition and Applications  
2014/05/05 Taipei Municipal Jianguo High School
  Title:  Recent Developments in Speech Recognition Technologies and their Applications  
2013/06/08 Workshop on Technology for Language Research and Language Learning, IACL 2013
  Title:  Speech Recognition and its Applications to Computer-Assisted Language Learning
2013/05/22 Graduate Institute of Communication Engineering, National Taiepi University (NTPU)
  Title:  Recent Developments in Language Modeling Techniques and their Applications
2013/01/11 Research Center for Psychological and Educational Testing (RCPET), National Taiwan Normal University (NTNU)
  Title:  Topic Language Models and their Applications
2013/01/03 Department of Chinese, National Taiwan Normal University (NTNU)
  Title:  Recent Developments in Speech Retrieval and Related Applications
2012/05/29 Graduate Institute of Library & Information Studies, National Taiwan Normal University (NTNU)
  Title:  Recent Developments in Speech Retrieval and Related Applications
2012/05/01 Department of Electronic Engineering, Ming Chi University of Technology (MCUT)
  Title:  Introduction to Automatic Speech Recognition and its Application
2011/12/28 CSIE Department, National Central University (NCU)
  Title:  Relevance Language Modeling for Speech Recognition and  Related Applications
2010/12/30 Department of Information Management, Yuan Ze University (YZU)
  Title:  Handling Verbose Queries for Speech Retrieval
2010/12/24 Department Computer Science and Engineering, Yuan Ze University (YZU)
  Title:    Language Modeling for Speech Recognition and Related Applications
2010/12/09 CS Department, National Cheng-chi University (NCCU)
  Title:    Speech summarization - From the view of decision theory
2010/08/11 Research Center for Psychological and Educational Testing (RCPET), National Taiwan Normal University (NTNU)
  Title:    Probabilistic Latent Topic Analysis and Its Applications
2010/05/14 CSIE Department, National Taiwan University (NTU)
  Title:    A risk minimization framework for extractive speech summarization
2010/05/04 2010 Speech Signal Processing Workshop (held at National Chiayi University)
  Title:    Recent Developments in Speech Retrieval and Summarization
2009/12/28 Graduate Institute of Library, Information and Archival Studies, National Cheng-chi University (NCCU)
  Title:    Spoken Document Recognition, Retrieval and Summarization
2009/11/24 EE Department, National Sun Yat-sen University (NSYSU)
  Title:    Recent Developments in Speech Recognition Technologies and Their Applications to Multimedia Information Access
  2009/10/02 EE Department, National Tsing Hua University (NTHU)
  Title:    Recent Developments in Speech Recognition Technologies and Their Applications to Multimedia Information Access
2009/05/04 CSIE Department, National Taiwan University of Science and Technology (NTUST)
  Title:    Latent topic modeling of word co-occurrence information for spoken document retrieval
2009/03/25 Research Center for Psychological and Educational Testing (RCPET), National Taiwan Normal University (NTNU)
  Title:    Speech Recognition Technologies and Their Applications to Multimedia Information Access
2009/01/21 Google Taipei
  Title: Recent Developments in Chinese Spoken Document Search and Distillation
2008/11/06 CSIE Department, National Chiao Tung University (NCTU)
  Title:    Spoken Document Retrieval and Summarization
2008/05/16 CSIE Department, National Sun Yat-sen University (NSYSU)
  Title: Exploiting Feature Distribution Characteristics for Robust Speech Recognition
2008/04/17 IM Department, National Taiwan University of Science and Technology (NTUST)
  Title: Spoken Document Retrieval and Summarization
2008/03/28 CS Department, National Defense University (NDU)
  Title: Spoken Document Retrieval and Summarization
2007/11/29 CS Department, National Cheng-chi University (NCCU)
  Title: Spoken Document Recognition, Retrieval and Summarization
2007/08/10 Compal Communications, Inc.
  Title: Speech Recognition Technology and Its Applications
2007/05/30 Telecommunication Laboratories, Chunghwa Telecom Co., Ltd.
  Title: Spoken Document Recognition, Organization and Retrieval
2007/05/07 Graduate Institute of Library & Information Studies, National Taiwan Normal University (NTNU)
  Title: Recent Progress in Spoken Document Recognition, Organization and Retrieval
2006/11/22 CSIE Department, National Central University (NCU)
  Title: Extractive Chinese Spoken Document Summarization Using Probabilistic Ranking Models
2006/08/03 National Institute of Information and Communications Technology (NiCT), ATR-SLC, Japan
  Title: Chinese Spoken Document Recognition, Organization and Retrieval
2006/06/08 EE Department, National Chi Nan University (NCNU)
  Title: Chinese Spoken Document Recognition, Organization and Retrieval
2006/03/20 Graduate Institute of Information and Computer Education, National Taiwan Normal University (NTNU)
  Title: Exploring Probabilistic Latent Semantic Information for Speech Processing
2006/01/13 Telecommunication Laboratories, Chunghwa Telecom Co., Ltd.
  Title: Data-Driven and Discriminative Approaches to Transcription, Retrieval and Organization of Mandarin Spoken Documents
2005/12/23 2005 Information Processing Workshop (held at Institute of Information Science, Academia Sinica)
  Title: Chinese Spoken Document Recognition, Organization and Retrieval
2005/11/14 CSIE Department, National Taiwan University of Science and Technology (NTUST)
  Title: Chinese Spoken Document Recognition, Retrieval and Organization
2005/08/12 Information & Communications Research Laboratories, Industrial Technology Research Institute (ITRI)
  Title: Recent Approaches to Transcription, Retrieval and Organization of Mandarin Broadcast News Speech
2005/05/04 Delta Electronics, Inc.
  Title: Recent Approaches to Transcription, Retrieval and Organization of Mandarin Broadcast News Speech
2004/11/18 EE Department, National Chi Nan University (NCNU)
  Title: Chinese Speech Recognition and Information Retrieval
2004/10/18 Department of Information and Computer Education, National Taiwan Normal University (NTNU)
  Title: Chinese Speech Recognition and Information Retrieval
2004/09/22 CSIE Department, National Taiwan Normal University (NTNU)
  Title: Chinese Speech Recognition and Information Retrieval
2004/04/26 Press Conference, National Museum of History
  Title: The development of automatic categorized indexing and multi-modal retrieval techniques for multimedia content of digital libraries
2004/04/09 CSIE Department, National Cheng Kung University (NCKU)
  Title: Lightly Supervised and Data-Driven Approaches to Mandarin Broadcast News Transcription
2003/12/18 2003 Workshop on Digital Archives Technology and Innovalue, National Taiwan Normal University (NTNU)
  Title: Voice Retrieval of Mandarin Speech Information
2003/11/13 VisionNext Co., Ltd.
  Title: Voice Retrieval of Mandarin Speech Information
2003/10/16 Graduate Institute of Biomedical informatics, Taipei Medical University
  Title: Chinese Speech Information Retrieval
2003/06/09 National Taipei University of Education (NTUE)
  Title: Speech Retrieval of Mandarin Broadcast News
2003/06/09 CSIE Department, National Tsing Hua University (NTHU)
  Title: Speech Retrieval of Mandarin Broadcast News
2002/04/26 2002 Speech Signal Processing Workshop
  Title: Retrieval of Mandarin Chinese Broadcast News via Speech
2001/07/23 Institute of Information Science, Academia Sinica
  Title: Speech Information Retrieval for Mandarin Chinese - Syllable- Based Indexing Features, Statistical Retrieval Models, Improved  Approaches 

- Others

Biographical profile included in the 25th Silver Anniversary Edition of Marquis Who's Who in the World, 2008
Biographical profile to be included in the 7th Edition of Marquis Who's Who in Medicine and Healthcare, 2009-2010